Complex microelectronic devices such as modern semiconductor chips require numerous connections to other electronic components. For example, a complex microprocessor chip may require many hundreds of connections to external devices.
Semiconductor chips commonly have been connected to electrical traces on mounting substrates by one of three methods: wire bonding, tape automated bonding, and flip-chip bonding. In wire bonding, the chip is positioned on a substrate with a bottom or back surface of the chip abutting the substrate and with the contact-bearing front or top surface of the chip facing upwardly, away from the substrate. Individual gold or aluminum wires are connected between the contacts on the chip and pads on the substrate. In tape automated bonding a flexible dielectric tape with a prefabricated array of leads thereon is positioned over the chip and substrate and the individual leads are bonded to the contacts on the chip and to pads on the substrate. In both wire bonding and conventional tape automated bonding, the pads on the substrate are arranged outside of the area covered by the chip, so that the wires or leads fan out from the chip to the surrounding pads. The area covered by the subassembly as a whole is considerably larger than the area covered by the chip. This makes the entire assembly substantially larger than it otherwise would be. Because the speed with which a microelectronic assembly can operate is inversely related to its size, this presents a serious drawback. Moreover, the wire bonding and tape automated bonding approaches are generally most workable with chips having contacts disposed in rows extending along the periphery of the chip. They generally do not lend themselves to use with chips having contacts disposed in a so-called area array, i.e., a grid-like pattern covering all or a substantial portion of the chip front surface.
In the flip-chip mounting technique, the contact bearing surface of the chip faces towards the substrate. Each contact on the chip is joined by a solder bond to the corresponding pad on the substrate, as by positioning solder balls on the substrate or chip, juxtaposing the chip with the substrate in the front-face-down orientation and momentarily melting or reflowing the solder. The flip-chip technique yields a compact assembly, which occupies an area of the substrate no larger than the area of the chip itself. However, flip-chip assemblies suffer from significant problems with thermal stress. The solder bonds between the chip contacts and substrate are substantially rigid. Changes in the size of the chip and of the substrate due to thermal expansion and contraction in service create substantial stresses in these rigid bonds, which in turn can lead to fatigue failure of the bonds. Moreover, it is difficult to test the chip before attaching it to the substrate, and hence difficult to maintain the required outgoing quality level in the finished assembly, particularly where the assembly includes numerous chips.
Numerous attempts have been made to solve the foregoing problem. Useful solutions are disclosed in commonly assigned U.S. Pat. Nos. 5,148,265 and 5,148,266. Preferred embodiments of the structures disclosed in these patents incorporate flexible, sheet-like structures referred to as "interposers" or "chip carriers". The preferred chip carriers have a plurality of terminals disposed on a flexible, sheet-like top layer. In use, the interposer is disposed on the front or contact bearing surface of the chip with the terminals facing upwardly, away from the chip. The terminals are then connected to the contacts of the chip. Most preferably, this connection is made by bonding prefabricated leads on the interposer to the chip contacts, using a tool engaged with the lead. The completed assembly is then connected to a substrate, as by bonding the terminals of the chip carrier to the substrate. Because the leads and the dielectric layer of the chip carrier are flexible, the terminals on the chip carrier can move relative to the contacts on the chip without imposing significant stresses on the bonds between the leads and the chip, or on the bonds between the terminals and the substrate. Thus, the assembly can compensate for thermal effects. Moreover, the assembly most preferably includes a compliant layer disposed between the terminals on the chip carrier and the face of the chip itself as, for example, an elastomeric layer incorporated in the chip carrier and disposed between the dielectric layer of the chip carrier and the chip. Such a compliant structure permits displacement of the individual terminals independently towards the chip. This permits effective engagement between the subassembly and a test fixture. Thus, a test fixture incorporating numerous electrical contacts can be engaged with all of the terminals in the subassembly despite minor variations in the height of the terminals. The subassembly can be tested before it is bonded to a substrate so as to provide a tested, known, good part to the substrate assembly operation. This in turn provides very substantial economic and quality advantages.
Commonly owned U.S. Pat. No. 5,455,390 describes a further improvement. Components according to preferred embodiments of the '390 patent use a flexible, dielectric top sheet having top and bottom surfaces. A plurality of terminals are mounted on the top sheet. A support layer is disposed underneath the top sheet, the support layer having a bottom surface remote from the top sheet. A plurality of electrically conductive, elongated leads are connected to the terminals on the top sheet and extend generally side by side downwardly from the terminals through the support layer. Each lead has a lower end at the bottom surface of the support layer. The lower ends of the leads have conductive bonding materials as, for example, eutectic bonding metals. The support layer surrounds and supports the leads.
Components of this type can be connected to microelectronic elements such as semiconductor chips or wafers by juxtaposing the bottom surface of the support layer with the contact-bearing surface of the chip so as to bring the lower ends of the leads into engagement with the contacts on the chip, and then subjecting the assembly to elevated temperature and pressure conditions. All of the lower ends of the leads bond to the contacts on the chip substantially simultaneously. The bonded leads connect the terminals of the top sheet with the contacts on the chip. The support layer desirably is either formed from a relatively low-modulus, compliant material, or else is removed and replaced after the lead bonding step with such a compliant material. In the finished assembly, the terminals desirably are movable with respect to the chip to permit testing and to compensate for thermal effects. However, the components and methods of the '390 patent provide further advantages, including the ability to make all of the bonds to the chip or other component in a single lamination-like process step. The components and methods of the '390 application are especially advantageous when used with chips or other microelectronic elements having contacts disposed in an area array.
Despite the positive results of the aforementioned commonly owned inventions, still further improvements would be desirable.